AMD Navi Product Reviews and Previews: (5500, 5600 XT, 5700, 5700 XT)

Yes. While there's mostly speculation at the start, more updates later in the thread start sourcing from LLVM and driver changes, and then the announcements from AMD. Within the last day or so, there's a link to a slide presentation outlining a decent amount of the CU and architectural modifications.
I read through the entire thread and wow, RDNA1 really is a departure from GCN. And that's only V1 of RDNA, imagine RDNA2 next year!
 
AMD Talks Radeon RX 5700 Temps And Max Boost on Their Blog
In what can only be described as an informative article, AMD uncovered some design features on the RX 5700 series in terms of internal temperatures, they are also saying that any RX 5700 is pushed close to its maximum performance 'so you don't have to tweak it'.

If you read through the blog post well, you'll find some interesting and even shocking points as apparently the Radeon RX 5700 XT can reach GPU "hotspot" temperatures passing 100 °C, actually hitting 113 °C with stress-testing, but that would be Furmark which we deem a viral on the GPU application. In its blog post, AMD mentions that 110 °C hotspot temperatures under "typical gaming usage" are "expected and within spec." I'm not sure that information will be good for their sales - though kudos for being open and transparent about it. AMD also talks "Maximized Performance, Right Out of the Box" and basically is saying that the card will go as far as it can right out of the box. I'll leave the paragraphs for you to digest, it will be interesting to see the comments"
https://www.guru3d.com/news-story/amd-talks-rx-5700-temps-and-max-boost-on-their-blog.html
 
I'm guessing this is the best thread to ask. Is there any way to find out whats different in Sapphire 5700 XT Pulse BIOSes when the company doesn't tell you? There's 4 different BIOSes, 2 of them .047 and 2 of them 0.49, only visible difference in what TPU BIOS database lists is slightly higher memory clock in one of the steps (in the middle so not affecting 3D clocks) on Samsung memories.
Wondering if I should update my .047 to the newest 0.49 or not
 
Computerbase.de: Sapphire Radeon RX 5700 XT Nitro+ vs RTX 2070 Super (FE): Overclocking, Undervolting and driver comparison tests
October 23, 2019
How do AMD Radeon RX 5700 (XT) and Nvidia GeForce RTX 2070 Super beat each other in terms of overclocking and undervolting? The editors have so far been guilty of answering due to time constraints. Now ComputerBase answers the outstanding questions and also takes a look at the advantage of PCIe 4.0 on Navi 10.
https://www.computerbase.de/2019-10...#diagramm-metro-exodus-2560-1440-undervoltage
 
I think from now on they'll just add features (RT, VRS) and not change the butterfly architecture, WGP and such.
The improvement should come from executing more integer ops per cycle (adding an integer unit?) to reach Nvidia gaming efficiency, right?. But they already have more transistors for a little less performance, so...
 
Last edited:
TPU review of the RX 5500 is up!

https://www.techpowerup.com/review/amd-radeon-rx-5500/

Looks like a solid mainstream gamer card, with much needed large improvements in efficiency over Polaris.

relative-performance_2560-1440.png


power-gaming-average.png
 
Yes the 2080Ti is hell of expensive, but damn the performance of that thing, it's a whole 90% faster then the 5700XT. It's the saturn V of the graphic cards :p
 
Yes the 2080Ti is hell of expensive, but damn the performance of that thing, it's a whole 90% faster then the 5700XT. It's the saturn V of the graphic cards :p

You are reading the chart wrong.

RTX 2080Ti is 2.9x faster than RX5500, whereas RX5700XT is 2x faster than R5500 at that resolution on those games.

RTX 2080Ti is in fact 45% faster than RX5700 XT on that chart.
 
Even with a node advantage, the perf/watt of RDNA is still hot trash. When are they going to solve that? If Navi was 12/16nm it’d be a furnace.
But it's not, so what's the point?

If your comparative metric is perf/watt then it is matching the competition regardless of node advantage and especially compared to the part it supersedes. We will probably find that AMD have yet again gone for a higher v-core to maintain desired yields and binning. The rest is down to different architecture.
 
I think we all want more competition, amd is on 7nm that nvidia hasn't moved to yet. That and AMD seems to only compete in the low/midrange segment.
 
As a general retort to those being less flattering towards the RDNA improvement compared ot the competition - as we've seen with Zen a lot of the developments planned weren't introduced immediately with the first Zen generation, some ended up introduced in Zen+, and yet more in Zen 2. I'm assuming the same goes for RDNA. It's quite evident that AMD isn't sitting on its hands, and with increased cash flow comes increased resources to all parts of the company. Even though Zen was premiered for the longest time it follows that all departments will be boosted. RDNA2 will probably be quite a refinement, benefitting from the groundwork laid by RDNA and the time given to prepare new parts for inclusion.
 
Back
Top